I might just stick with jq instead of adding another thing to maintain.
EverythingSings
EverythingSings@primal.net
npub1cdm4...er54
Formless art for the future
Currently I'm exploring if the #SCT infrastructure would benefit from a more formal interface or communication standard. I don't want to over-engineer (too much), it's possible my JSON flags are enough.
With #SCT, I'm trying to design the architecture from the perspective 'what tools would not only benefit me, but also the AI agents I work with.'
‘Shell’
Ink: Sharpie Pro Chisel Permanent Marker
Paper: gry mattr by joe mimran, DARE TO IMAGINE EVERYTHING
#art #drawing
Sleeping in after Thanksgiving is a time honored tradition
studying simplicity
I also want to make it so that growth is guaranteed, each rule should expand the network in some way. I don't want to explore a bunch of networks that remain one node for 1000 ticks.
I'm simplifying the model, or trying to, so it's not thousands of rules to explore.
I spent all the money I want to at the moment but I am making interesting progress.
I'm having a discussion with Claude Opus 4.5 from an architectural level and then going back to Shakespeare and implementing plans with Sonnet 4.5.
Wolfram exhaustively explored 256 deterministic rules on a fixed 1D grid to discover that minimal systems can produce maximal complexity; this project asks the same question but for a dynamic relational substrate. 5,760 rules where nodes spawn, die, and connect, probability is fundamental, and rules can sense their own history.
Claude Sonnet 4.5 loves my master prompt: This is beautiful. Clean, finite, exhaustive, and actually runnable. Let me build this.
Trying out Shakespeare by @Alex Gleason
Small privacy tip if you use OpenRouter as your provider:
Settings -> Training, Logging, and Privacy -> Disable Everything except ZDR Endpoints Only
We can already see the rarity distribution of the #Lumimenta cards, but something I might add is a 'release schedule', or a projection of how many cards will be in circulation at X date. That way I can track if I need to print more or slow down.
The SWC-62 council will rely on the #SCT infrastructure.
The current idea is that there will be 6 members on the council. Each council member will focus on a specific sector: quantum information science, advanced manufacturing, biotechnology, critical materials, nuclear fission/fusion energy, semiconductors and microelectronics.
The codename for the Council of LLMs project is SWC-62.
I would say the #SystemicWisdom experiment was a success, because it had enough potential energy to briefly capture the attention of a hive mind.
The experiment is paused because I didn't receive much support (it does cost money) and because I want to explore a Council of LLMs next, instead of just one RAG system.
One of my experiments this year was an AI agent that thought to itself over an extended period of time. A retrieval augmented generation system, except it's only augmented by its own thinking, and not a knowledge base. We call it SystemicWisdom. I'm pausing the experiment for a bit.